In this paper, we propose a dictionary update method for Nonnegative Matrix Factorization (NMF) with high dimensional data in a spectral conversion (SC) task. Voice conversion has been widely studied due to its potential applications such as personalized speech synthesis and speech enhancement. Exemplar-based NMF (ENMF) emerges as an effective and probably the simplest choice among all techniques for SC, as long as a source-target parallel speech corpus is given. ENMF-based SC systems usually need a large amount of bases (exemplars) to ensure the quality of the converted speech. However, a small and effective dictionary is desirable but hard to obtain via dictionary update, in particular when high-dimensional features such as STRAIGHT spectra are used. Therefore, we propose a dictionary update framework for NMF by means of an encoder-decoder reformulation. Regarding NMF as an encoder-decoder network makes it possible to exploit the whole parallel corpus more effectively and efficiently when applied to SC. Our experiments demonstrate significant gains of the proposed system with small dictionaries over conventional ENMF-based systems with dictionaries of same or much larger size.

We propose a novel kernel based post selection inference (PSI) algorithm, which can not only handle non-linearity in data but also structured output such as multi-dimensional and multi-label outputs. Specifically, we develop a PSI algorithm for independence measures, and propose the Hilbert-Schmidt Independence Criterion (HSIC) based PSI algorithm (hsicInf). The novelty of the proposed algorithm is that it can handle non-linearity and/or structured data through kernels. Namely, the proposed algorithm can be used for wider range of applications including nonlinear multi-class classification and multi-variate regressions, while existing PSI algorithms cannot handle them. Through synthetic experiments, we show that the proposed approach can find a set of statistically significant features for both regression and classification problems. Moreover, we apply the hsicInf algorithm to a real-world data, and show that hsicInf can successfully identify important features.

We introduce the localized Lasso, which is suited for learning models that are both interpretable and have a high predictive power in problems with high dimensionality $d$ and small sample size $n$. More specifically, we consider a function defined by local sparse models, one at each data point. We introduce sample-wise network regularization to borrow strength across the models, and sample-wise exclusive group sparsity (a.k.a., $\ell_{1,2}$ norm) to introduce diversity into the choice of feature sets in the local models. The local models are interpretable in terms of similarity of their sparsity patterns. The cost function is convex, and thus has a globally optimal solution. Moreover, we propose a simple yet efficient iterative least-squares based optimization procedure for the localized Lasso, which does not need a tuning parameter, and is guaranteed to converge to a globally optimal solution. The solution is empirically shown to outperform alternatives for both simulated and genomic personalized medicine data.

Beijing National Lab for Condensed Matter Physics and Institute of Physics, Chinese Academy of Sciences, Beijing 100190, China Despite their exceptional flexibility and popularity, the Monte Carlo methods often suffer from slow mixing times for challenging statistical physics problems. We present a general strategy to overcome this difficulty by adopting ideas and techniques from the machine learning community. We fit the unnormalized probability of the physical model to a feedforward neural network and reinterpret the architecture as a restricted Boltzmann machine. Then, exploiting its feature detection ability, we utilize the restricted Boltzmann machine for efficient Monte Carlo updates and to speed up the simulation of the original physical system. We implement these ideas for the Falicov-Kimball model and demonstrate improved acceptance ratio and autocorrelation time near the phase transition point. Monte Carlo method is one of the most flexible and powerful methods for studying many-body systems [1, 2]. Monte Carlo methods randomly sample configurations and obtain the answer as a statistical average.

The Founder and Executive Chairman of the World Economic Forum published a book in January called "The 4th Industrial Revolution." In his book, Klaus Schwab proposes that we are at the beginning of a fundamental shift in human/technological interaction. Advances in robotics will redefine the workforce. Advances in information technology will deliver meaningful artificial intelligence. And advances in biotechnology will redefine what it means to be human.

Recently, Microsoft took the unusual step of placing its Cortana and Bing product teams inside the same organization as Microsoft Research. The new Microsoft AI and Research Group will be led by computer vision pioneer and executive vice president Harry Shum, whose 20-year Microsoft career involves leading Bing's search efforts from 2007 through 2013 and helping launch Microsoft Research China. We asked Shum how this new organization will benefit Microsoft's digital assistant in the following interview, which has been edited for length and clarity. The language of the blog post announcing the formation of Microsoft's new AI division, together with how Satya Nadella has characterized it, suggests that Microsoft thinks it's in a space race of sorts when it comes to artificial intelligence. I just feel that the timing's right to go big on AI.

FILE- In this March 1, 2013 file photo, anti-Syrian President Bashar Assad protesters hold the Jabhat al-Nusra flag, as they shout slogans during a demonstration, at Kafranbel town, in Idlib province, northern Syria. Insurgent groups like Hezbollah and the Islamic State group in Syria have learned how to weaponize surveillance drones and use them against each other, adding a new twist to the country's civil war, a U.S. military official and others say. FILE- In this March 1, 2013 file photo, anti-Syrian President Bashar Assad protesters hold the Jabhat al-Nusra flag, as they shout slogans during a demonstration, at Kafranbel town, in Idlib province, northern Syria. Insurgent groups like Hezbollah and the Islamic State group in Syria have learned how to weaponize surveillance drones and use them against each other, adding a new twist to the country's civil war, a U.S. military official and others say. WASHINGTON (AP) -- French and Kurdish forces in northern Iraq were attacked by an exploding drone, the Pentagon said Wednesday, adding a new worry to the wars in Iraq and Syria as militant groups learn to weaponize their store-bought drones.

Just one day after announcing his plans to get humans to Mars by the 2030s, President Obama's thoughts on artificial intelligence have been revealed. According to the president, AI has been'seeping into our lives' for some time, even in ways we don't notice – and while there are many potential benefits that come with the technology, there may be grim consequences as well. The rise of'specialized AI' could deepen income inequality and make low-skilled employees become redundant in the workforce, Obama warned in an interview with Wired, and advanced, self-teaching algorithms could pose threats to the stock exchange and national security. According to the president, AI has been'seeping into our lives' for some time, even in ways we don't notice – and while there are many potential benefits that come with the technology, there may be grim consequences as well The White House has released a report and strategic plan regarding future directions and considerations for AI. The report, entitled Preparing for the Future of Artificial Intelligence, examines the current state of AI, along with its applications both now and in the future, and outlines a series of recommendations to ensure safe use.

Baidu, a China-based search engine business, took the wraps off a digital health tool to field medical queries and conversations between physicians and their patients called Melody medical assistant. The company claimed in a news release that the app uses deep learning to help doctors gather information from patients about their medical conditions and help physicians arrive at a diagnosis. To give an idea how the bot is designed to work, a spokeswoman provided an overview, in response to emailed questions. When a patient opens the app to pose a question, Melody asks the patient relevant follow-up questions to clarify information such as the duration, severity, and frequency of symptoms. The questions can also touch on additional symptoms related to the condition, even though the patient may not have mentioned them. The point is to give the doctor a more detailed sense of the patient's condition to decide whether to recommend the patient for an appointment sooner rather than later.

Are you looking for some new Xbox One games to play? We've compiled a list of the top 10 best upcoming Xbox One games for 2016. Gears of War 4 will be released for Xbox One on October 11, 2016. Gears of War 4 takes place 25 years after the events of Gears of War 3, and follows JD Fenix and his friends, Kait and Del, on a mission to rescue their loved ones and discover the source of a new enemy. Fans can expect familiar gameplay, multiplayer and stunning visuals by Unreal Engine 4. Batman: Return to Arkham features remastered versions of the first two games in the Batman: Arkham series, including Batman: Arkham Asylum and Batman: Arkham City.